US9100053B2ActiveUtilityPatentIndex 51
Method and decoder for reconstructing a source signal
Est. expiryMay 23, 2031(~4.9 yrs left)· nominal 20-yr term from priority
H03M 7/40H03M 13/29H03M 7/3059H04B 1/04H04N 19/44H03M 7/4006H04N 19/30
51
PatentIndex Score
1
Cited by
24
References
20
Claims
Abstract
In a method for reconstructing a source signal, which is encoded by a set of at least two descriptions, the method comprises: receiving a subset of the set of descriptions; reconstructing a reconstructed signal at an operating bitrate of a set of operating bitrates upon the basis of the subset of descriptions, the reconstructed signal having a second probability density, wherein the second probability density comprises a first statistical moment and a second statistical moment; and manipulating the reconstructed signal, wherein the reconstructed signal is manipulated such that, irrespective of the operating bitrate, a predetermined minimum similarity between the first statistical moment of the third probability density and the first statistical moment of the first probability density and between the second statistical moment of the third probability density and the second statistical moment of the first probability density is maintained.
Claims
exact text as granted — not AI-modifiedWhat is claimed is:
1. A method for reconstructing a source signal, which is encoded by a set of at least two descriptions, the source signal having a first probability density, wherein the first probability density comprises a first statistical moment and a second statistical moment, the method comprising:
receiving a subset of the set of descriptions;
reconstructing a reconstructed signal at an operating bitrate of a set of operating bitrates upon the basis of the subset of descriptions, the reconstructed signal having a second probability density, wherein the second probability density comprises a first statistical moment and a second statistical moment; and
manipulating the reconstructed signal in order to obtain a manipulated reconstructed signal having a third probability density,
wherein the third probability density comprises a first statistical moment and a second statistical moment,
wherein the reconstructed signal is manipulated such that at least the first statistical moment and the second statistical moment of the third probability density are more similar to the first statistical moment and the second statistical moment of the source signal than the first statistical moment and the second statistical moment of the second probability density are, and
wherein the reconstructed signal is manipulated such that, irrespective of the operating bitrate, a predetermined minimum similarity between the first statistical moment of the third probability density and the first statistical moment of the first probability density and between the second statistical moment of the third probability density and the second statistical moment of the first probability density is maintained.
2. The method according to claim 1 , wherein the first statistical moment and the second statistical moment of the second probability density are manipulated to preserve the first statistical moment and the second statistical moment of the first probability density within a predetermined moment range.
3. The method according to claim 1 , wherein the first statistical moments of the first probability density and the third probability density are about equal, and wherein the second statistical moments of the first probability density and the third probability density are about equal.
4. The method according to claim 1 , wherein the reconstructed signal is reconstructed using a reconstruction function that is dependent on a composition of descriptions in the subset of descriptions.
5. The method according to claim 1 , wherein the source signal comprises an additive dither signal, and wherein the reconstructing comprises subtracting the dither signal from the reconstructed signal.
6. The method according to claim 1 , wherein the source signal comprises a pseudorandom dither signal, and wherein the reconstructing comprises subtracting the pseudorandom dither signal from the reconstructed signal.
7. The method according to claim 1 , wherein reconstructing comprises using an index assignment scheme, which is addressed by the descriptions of the set of descriptions, the index assignment scheme being used for deriving the set of descriptions encoding the source signal.
8. The method according to claim 1 , wherein manipulating the reconstructed signal comprises transforming the reconstructed signal according to a statistical transformation function, the transformation function being dependent on a composition of descriptions in the subset of descriptions.
9. The method according to claim 1 , wherein manipulating the reconstructed signal comprises transforming the reconstructed signal according to a statistical transformation function T(x), the transformation function T(x) being defined according to the following formula:
T
(
x
)
=
F
X
-
1
{
1
Δ
∫
Δ
2
Δ
2
F
X
(
x
-
τ
)
ⅆ
τ
}
,
where Δ is a quantizer step size, F X (x) is the cumulative distribution function of variable X that is related to the probability density function ƒ X (·) of the first probability density, as
F
X
(
x
)
=
∫
-
∞
x
f
X
(
τ
)
ⅆ
τ
,
and F X −1 (·) denotes the inverse cumulative distribution function.
10. The method according to claim 1 , wherein manipulating the reconstructed signal comprises transforming the reconstructed signal according to a statistical transformation function T(x), the transformation function T(x) being defined according to the following formula:
T
(
x
)
=
σ
X
2
σ
X
2
+
Δ
2
12
x
,
where Δ is a quantizer step size, and
σ
X
2
=
∫
-
∞
∞
x
2
f
X
(
x
)
ⅆ
x
is the variance of variable X that is related to the probability density function ƒ X (·) of the first probability density.
11. The method according to claim 1 , wherein manipulating the reconstructed signal comprises transforming the reconstructed signal according to a statistical transformation function T(x), the transformation function T(x) being defined according to the following formula:
T
(
x
)
=
F
X
-
1
{
1
2
M
Δ
∑
i
∈
P
(
M
)
∫
Δ
l
(
i
)
Δ
r
(
i
)
F
X
(
x
-
τ
)
ⅆ
τ
}
,
where Δ is a quantizer step size, M is an index assignment parameter, F X (x) is the cumulative distribution function of variable X that is related to the probability density function ƒ X (·) of the first probability density, as
F
X
(
x
)
=
∫
-
∞
x
f
X
(
τ
)
ⅆ
τ
,
F X −1 (·) denotes the inverse cumulative distribution function, and P(M) is an index assignment pattern of an index assignment scheme being used for deriving the set of descriptions encoding the source signal,
Δ
l
(
i
)
=
-
(
i
-
P
_
)
Δ
-
Δ
2
and
Δ
r
(
i
)
=
-
(
i
-
P
_
)
Δ
+
Δ
2
,
with
P
_
=
1
2
M
∑
i
∈
P
(
M
)
i
.
12. The method according to claim 1 , wherein manipulating the reconstructed signal comprises transforming the reconstructed signal according to a statistical transformation function T(x), the transformation function T(x) being defined according to the following formula:
T
(
x
)
=
σ
X
2
σ
X
2
+
Δ
2
M
4
3
x
,
where Δ is a quantizer step size, M is an index assignment parameter and
σ
X
2
=
∫
-
∞
∞
x
2
f
X
(
x
)
ⅆ
x
is the variance of variable X that is related to the probability density function ƒ X (·) of the first probability density.
13. A decoder for reconstructing a source signal, which is encoded by a set of at least two descriptions, the source signal having a first probability density, wherein the first probability density comprises a first statistical moment and a second statistical moment, the decoder comprising:
an input for receiving a subset of the set of descriptions;
a reconstructor for reconstructing a reconstructed signal at an operating bitrate of a set of operating bitrates upon the basis of the subset of descriptions, the reconstructed signal having a second probability density, wherein the second probability density comprises a first statistical moment and a second statistical moment; and
a transformer for manipulating the reconstructed signal in order to obtain a manipulated reconstructed signal having a third probability density,
wherein the third probability density comprises a first statistical moment and a second statistical moment,
wherein the transformer is configured to manipulate the reconstructed signal such that at least the first statistical moment and the second statistical moment of the third probability density are more similar to the first statistical moment and the second statistical moment of the source signal than the first statistical moment and the second statistical moment of the second probability density are, and
wherein the transformer is furthermore configured to manipulate the reconstructed signal such that, irrespective of the operating bitrate, a predetermined minimum similarity between the first statistical moment of the third probability density and the first statistical moment of the first probability density and between the second statistical moment of the third probability density and the second statistical moment of the first probability density is maintained.
14. The decoder of claim 13 , wherein the transformer is configured to manipulate the first statistical moment and the second statistical moment of the second probability density in order to preserve the first statistical moment and the second statistical moment of the first probability density within a predetermined moment range.
15. The decoder of claim 13 , wherein the reconstructor comprises a central reconstruction path, which is configured to reconstruct the reconstructed signal upon the basis of index information, the central reconstruction path comprising an indexer configured to determine the index information upon the basis of the set of descriptions.
16. The decoder of claim 13 , wherein the reconstructor comprises a central reconstruction path, which is configured to reconstruct the reconstructed signal upon the basis of unique index information, the central reconstruction path comprising an indexer configured to determine the unique index information upon the basis of the set of descriptions.
17. The decoder of claim 13 , wherein the reconstructor comprises at least one side reconstruction path, which is configured to reconstruct the reconstructed signal upon the basis of mapping information, the at least one side reconstruction path comprising a mapper configured to determine the mapping information upon the basis of the descriptions of the subset and of a composition of descriptions in the subset.
18. The decoder of claim 13 , wherein the transformer is configured to perform the manipulating upon the basis of a composition of descriptions in the subset of descriptions.
19. At least one processor configured to reconstruct a source signal, which is encoded by a set of at least two descriptions, the source signal having a first probability density, wherein the first probability density comprises a first statistical moment and a second statistical moment, by:
receiving a subset of the set of descriptions;
reconstructing a reconstructed signal at an operating bitrate of a set of operating bitrates upon the basis of the subset of descriptions, the reconstructed signal having a second probability density, wherein the second probability density comprises a first statistical moment and a second statistical moment; and
manipulating the reconstructed signal in order to obtain a manipulated reconstructed signal having a third probability density,
wherein the third probability density comprises a first statistical moment and a second statistical moment,
wherein the reconstructed signal is manipulated such that at least the first statistical moment and the second statistical moment of the third probability density are more similar to the first statistical moment and the second statistical moment of the source signal than the first statistical moment and the second statistical moment of the second probability density are, and
wherein the reconstructed signal is manipulated such that, irrespective of the operating bitrate, a predetermined minimum similarity between the first statistical moment of the third probability density and the first statistical moment of the first probability density and between the second statistical moment of the third probability density and the second statistical moment of the first probability density is maintained.
20. The processor according to claim 19 , wherein the first statistical moment and the second statistical moment of the second probability density are manipulated to preserve the first statistical moment and the second statistical moment of the first probability density within a predetermined moment range.Cited by (0)
No later patents cite this yet.
References (0)
No backward citations on record.